What happened
Anthropic's Claude services went down Sunday evening in an authentication outage that prevented users from logging in across claude.ai, the Claude API, Claude Code, Claude Cowork, and the platform.claude.com developer console. The company's status page logged the first notice at 21:58 UTC, which is 5:58 p.m. Eastern, and widened the incident minutes later to cover degraded performance reports. Outage trackers collected more than 15,000 error reports before service was restored, with the status page marking recovery around 22:34 UTC, roughly 40 minutes after it began. Anthropic has not publicly disclosed a root cause. It was the company's second multi-service disruption in about two weeks, following network failures that cut Claude capacity on July 29 and 30.
Why it matters for your business
A 40-minute Sunday outage is a minor event on its own. The reason to pay attention is what it reveals: AI assistants have quietly become production infrastructure for a lot of small businesses, used for drafting, coding, customer responses, and research, and most of those businesses have no plan for when the tool is unavailable. If a chunk of your team's daily work now routes through one AI service, its uptime is your uptime, the same as email or your phone system.
What to do about it
- Know where your AI provider publishes status updates so an outage costs you two minutes of checking instead of an hour of confusion
- For business-critical workflows built on one AI service, identify a second tool or a manual fallback before you need it
- Keep important prompts, documents, and outputs saved outside the chat interface so an outage does not lock up your work product